TDLAS系统浓度反演方法、装置、设备、存储介质和程序产品
By combining the AEEMAM-DCM neural network with CEEMDAN decomposition and channel attention mechanism, noise and effective signals are adaptively separated. Deformable convolution is used to optimize feature extraction, which solves the problems of weak noise resistance and limited receptive field of the TDLAS system and achieves high-precision inversion of gas concentration and temperature.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA NAT PETROLEUM CORP
- Filing Date
- 2024-09-06
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional TDLAS systems have weak noise immunity, and convolutional neural networks lack a mechanism for separating noise from effective signals, resulting in a limited receptive field and low accuracy in gas concentration retrieval.
We employ an AEEMAM-DCM neural network, combined with CEEMDAN decomposition and channel attention mechanism, to adaptively separate noise from effective signals. We use deformable convolution to increase the receptive field and add a positional attention mechanism to optimize feature extraction.
It improves the accuracy and noise resistance of gas concentration inversion, realizes high-precision inversion of gas concentration and temperature, and enhances the signal-to-noise ratio and sensitivity of the system.
Smart Images
Figure CN121637050B_ABST